Skills
skills/nvidia/skills/tao-train-mask-auto-encoder/Gen Agent Trust Hub

tao-train-mask-auto-encoder

Pass

Audited by Gen Agent Trust Hub on Jun 23, 2026

Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONDATA_EXFILTRATION
Full Analysis
  • [SAFE]: The skill is an official contribution from NVIDIA and includes digital signatures (skill.oms.sig) and integrity manifests to ensure the authenticity of its files.
  • [EXTERNAL_DOWNLOADS]: The skill documentation and references point to official Docker images hosted on the NVIDIA Container Registry (nvcr.io). These are trusted, well-known sources for machine learning workflows.
  • [COMMAND_EXECUTION]: The skill defines command patterns for the mae CLI tool. These operations are intended for use within a containerized TAO Toolkit environment and align with the skill's stated purpose.
  • [DATA_EXFILTRATION]: Schemas include optional integration with Weights & Biases (wandb) for standard experiment tracking. No unauthorized data exfiltration or hardcoded credentials were detected.
  • [SAFE]: The use of Base64 encoding in the skill.oms.sig file is part of a standard digital signature bundle format (DSSE) and is used for integrity verification rather than malicious obfuscation.
Audit Metadata
Risk Level
SAFE
Analyzed
Jun 23, 2026, 12:10 PM
Security Audit — agent-trust-hub — tao-train-mask-auto-encoder